Molecular Modeling - American Chemical Society

Chapter 17

Molecular Dynamics and NMR Studies of Concentrated Electrolytes and Dipoles in Water 1

1

1

2

Ion C. Baianu , E. M . Ozu , T. C. Wei , and Thomas F. Kumosinski

Downloaded by MONASH UNIV on October 26, 2012 | http://pubs.acs.org Publication Date: December 14, 1994 | doi: 10.1021/bk-1994-0576.ch017

1

Department of Food Science, Agricultural and Food Chemistry—Nuclear Magnetic Resonance Facility, University of Illinois at Urbana, 580 Bevier Hall, 905 South Goodwin Avenue, Urbana, IL 61801 Eastern Regional Research Center, Agricultural Research Service, U.S. Department of Agriculture, 600 East Mermaid Lane, Philadelphia, PA 19118 2

The molecular dynamics of water and selected ions was studied in concentrated electrolyte solutions with, or without, dipolar ions added. Our experimental results by multinuclear spin relaxation techniques were then compared with molecular dynamics computations for water and ions in concentrated electrolyte solutions (LiCl•R(H O)/R(D O) and NaCl•R(H O)/ 2

2

2

+

R(D O), with 4 < R < 12 for Li and 6 < R < 16 for NaCl). Multinuclear spin relaxation data were analyzed with a thermodynamic linkage model of hydrated ion clusters of various sizes and composition. Our results indicate that tetramer clusters of hydrated Li and Cl are the preferred structures formed in such concentrated electrolyte solutions as a consequence of dimertetramer equilibria that occur at 293 K. Within such clusters water molecules undergo hindered reorientation motions in the hydration shell of the cation. The corresponding correlation time of water (D O), determined by ONMR, is less than 30 ps for 4 < R < 12 in all solutions studied at 293 K. 2

+

-

17

2

The study of the dynamics of molecules by computer modeling is a relatively recent and rapidly expanding field of research; this field is now known as "Molecular Dynamics". A standard approach to molecular dynamics computations on a fast computer or a supercomputer begins with an arbitrary, or pseudo-random, configuration of a group of molecules or 'particles', and computes their subsequent positions and velocities as a function of time with Newton's equations of motions for these particles. The forces acting on molecules are derived from potential functions for a large number of particles in a "box" which is also set up with a grid for convenience in following the particle configurations and their statistics. Because of the complexity of the potential functions (interaction potentials), the reliability of the method is greatest for the simplest systems, such as "inert" gases and molten salts (1,2), (Figure la). Extensive molecular dynamics computations were also carried out for liquid water and aqueous solutions of electrolytes (2). In the latter case, the interaction potentials are not really known but they were assumed to involve only slight perturbations of the potentials for liquid water, which in itself is an approximation. For aqueous solutions of electrolytes with ionic radii that are larger than about 0.9 Â, and single charge, this approach might produce results that resemble 0097-6156/94/0576-0269$14.48/0

© 1994 American Chemical Society In Molecular Modeling; Kumosinski, T., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 1994.

270

MOLECULAR MODELING

the local structure of electrolyte solutions determined experimentally. For electrolytes with high ionic field strengths, such as I i , Ca , Zn , Be , L a , F", OH", etc, the results of this molecular dynamics approach disagree with local structures (Figure lb) that are emergingfromexperimental studies (3-5). Compare, for example, the local structure of molten KC1 in Figure la (which was derived from molecular dynamics computations) with that in Color Plate 15 for concentrated LiCl solutions in water (which was derived from Nuclear Magnetic Resonance (NMR) studies). The local structures of KC1, NaCl and CsCl in aqueous solutions derived from molecular dynamics computations were all similar to that shown in Figure la for molten KC1. Clearly, the local structures (4) of IiCl-nH20 (2.2 £ η £ 12) solutions shown in Color Plate 15 and Figures lb and lc are quite different from the proposed molecular dynamics structures for the same system. Certain experimental results that are available for liCl-idÔ (3-5) indicate that the Downloaded by MONASH UNIV on October 26, 2012 | http://pubs.acs.org Publication Date: December 14, 1994 | doi: 10.1021/bk-1994-0576.ch017

+

2+

2+

2+

+

3+

+

high-ionic field strength of L i causes water-bridging between L i and CI", as well as the formation of hydrated ion-pair clusters (Figure lb and Color Plate 16), whose presence and structure could not be detected in the molecular dynamics results (2).

Monte-Carlo Simulations and Spectroscopy A more recent approach to the molecular dynamics of aqueous solutions of electrolytes (6) involved the use of statistical mechanics combined with Monte Carlo simulations. The interparticle potential and the field gradient function employed to calculate thefluctuationsin solutions were based upon quantum mechanics/quantum chemistry calculations (6). This approach allows one to estimate the correlation times related to the computed fluctuations in solution. The Monte Carlo simulation involves the generation of a large number of molecule/particle configurations, starting from an equilibrium ensemble; the total potential energy is then assumed to be reasonably approximated by a sum of pair-interaction energies. One proceeds then to compute any physical property that can be expressed as a function of the interparticle distances and orientations. Such configurations generated in the Monte Carlo simulation on a mainframe computer (or, preferably, on a supercomputer) can be also employed to calculate hydration numbers, theoretical radial distribution functions, or other interesting properties of electrolyte solutions. The main advantage of this approach is that one can compares directly the predictions of molecular dynamics computations with the analysis of experimental results. For example, one can compare molecular dynamics results with spectroscopic (NMR, IR, Raman, etc) and X-ray/neutron/ electron scattering data for electrolyte solutions. Among the spectroscopic techniques widely used in studies of electrolyte solutions most prominent are laser-Raman scattering and Nuclear Magnetic Resonance (NMR). Nuclear Magnetic Resonance has the advantages of being able to monitor all the nuclei present in aqueous solutions of electrolytes, provide directly dynamic information through relaxation studies and allow one to derive hydration numbers from spectral (high-resolution NMR) studies/measurements of chemical shifts. Since nuclear spin relaxation studies are often able to estimate correlationtimesfor solutions by making only a minimum number of assumptions, such results are especially relevant to the evaluation of molecular dynamics computations.

Nuclear Magnetic Resonance Spectroscopy and Relaxation Nuclear Magnetic Resonance (NMR) is a branch of radiofrequency(r.f.) absorption spectroscopy which is specifically concerned with the resonant absorption of radiowaves by the nuclei of a sample placed in an intense magnetic field; the NOTE: The color plates can be found in a color section in the center of this volume.

In Molecular Modeling; Kumosinski, T., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 1994.

17.

BAIANU ET AL.

Concentrated Electrolytes and Dipoles in Water

271

radiofrequency absorption occurs as a result of transitions between the nuclear spin energy levels, in the presence of a very homogeneous, strong and static magnetic field. The stronger the magnetic field, Ho, the higher is the frequency required for resonance, and the more intense is the absorption, as indicated in Figure 2. The transition induced between the nuclear spin energy levels (Figure 2) by the resonant r.f. wave or pulse is recorded as a sharp peak for a liquid; for a pair of nuclear spin energy levels shown in Figure 2, a single peak is recorded whose lineshape is Lorentzian in a simple liquid or solution. The linewidth of the NMR absorption peak at half-height is determined by the lifetime of the excited nuclear spin state in the presence of only negligible magnetic field inhomogeneities. More precisely expressed, after the occurrence of the NMR absorption the system of nuclear spins relaxes through interactions between the nuclear spins ('spin-spin*, or T relaxation process). The NMR signal is recorded with a coil whose axis, x, is perpendicular to the direction ζ of the static magnetic field, Ho, in Figure 2; therefore, the loss of phase coherence of the nuclear spins in the xy-plane, which is normal to Η©, occurs as a result of transverse (T2) nuclear spin relaxation processes (Figure 3a) that involve interactions between nuclear spins. The corresponding Free Induction Decay (FID) signal is shown in Figure 3b. On the other hand, the nuclear spin magnetization has a component M along the magnetic field direction (z), which


2

z

is not directly observed by the detector coil whose axis is in the xy-plane, along the xdirection. Immediately after the nuclear spin excitation, the M component points z

against the static magnetic field H© and later relaxes (or comes back) with a characteristic time constant, Tj, towards the magnetic field direction. This relaxation process which occurs along the z-axis is called the longitudinal, or 'spin-lattice* (Ti) relaxation. The latter name is often used because the Ti-relaxation process involves interactions of the nuclear spins with the surrounding electrons ("the lattice"); in a crystalline solid, the nuclear spins interact with the electrons in the surrounding crystal lattice causing the relaxation of the M , magnetization component towards the magnetic field direction. The use of the term 'spin-lattice* relaxation is, however, not restricted to Ti -relaxation in crystalline solids but is employed, in general, for any system, either solid, liquid or gas. Mechanisms of Relaxation in a Liquid A. For_ajspjn 1=1/2, (such as *H) the major contributions to relaxation are made by: - spin-spin coupling - magnetic dipolar interactions - chemical exchange (of protons). B. For quadrupolar nuclei — (with spin I > 1/2, such as 0), the relaxation is caused by the interactions of the nuclear quadrupole moment with the surrounding, fluctuating electrical field. In the case of deuterium ( H) NMR, however, chemical exchange also contributes to the nuclear spin relaxation. Nuclear spin relaxation is, therefore, dependent upon the molecular dynamics of the liquid, which is characterized by a correlation time, or distribution of correlation times: the faster the motions are, the shorter is the correlation time and the longer is the relaxation time. A motion of a molecule, such as a rotation around a specific 17

2



272

MOLECULAR MODELING

Figure la. Computer generated configuration of ions in molten potassium chloride, based on a molecular dynamics program written in FORTRAN-77. (Modified from réf. 1).

+

Figure lb. Local structure of water bridged Li (4H 0)C1 clusters in glasses at 100 K, derived from pulsed H NMR data. (Modifiedfromref. 5). 2

l



17.

BAIANU ET AL.


273

Figure lc. Schematic model of the local structure of aqueous solutions of LiCl containing hydrated ion-pairs in Li (nH20)Cr, water-bridged clusters. +

D bond vector can be characterized by one to three order parameters, depending on the types of partial disorder encountered. For D2O molecules in a lattice, the variation in the orientation of the D2O molecule symmetry axis throughout the lattice can be represented by a single order parameter, S; the reduction in the quadrupolar splitting caused by such static disorder can be calculated from a powder spectrum with the following equation: 2

Av = (3/4).(e qQ/fe)S.T q

(2)

C

where Av is the residual quadrupolar spUrting and T is the average correlation time q

c

of the D2O molecules in the partially disordered lattice. For a completely random, "amorphous" solid, S = 0, whereas for a "perfect" crystal lattice, with no variation in the orientation of D2O molecules, S = 1.0. It is interesting that even for relatively small values of S, the deuterium NMR powder patterns retain sharp features in the absence of fast motions (8), if the dipolar interactions involving the deuterons are not strong. Deuterium NMR lineshapes are also markedly affected by exchange processes, such as the rotational jumps or "flips" of deuterons amongst alternate sites. An example is shown in Figure 6 where the H NMR lineshapes are presented as a function of the exchange rate (or "flip rate") of deuterons in an aromatic ring system (10) (e qQ/h =180 kHz and η = 0.06). Large-angle flips yield unique "double-horn" features (Figure 6B) when these occur about an axis which is itself moving. Such a dynamic model is intuitively appealing for aromatic rings trapped within channel clathrates (or embedded in highly ordered smectic mesophases) and for phenyl group substituents in proteins. This type of model can be generalized to include more orientations around the flip axis to simulate the motions of adsorbates trapped near surfaces or on catalytically active sites. 2

2

Local Structure and Molecular Dynamics in Aqueous Solutions of Electrolytes X-ray and Neutron Scattering Studies of Local Structure in Aqueous Solutions of Electrolytes. Understanding the local, or short-range (r < 5 Â), structure of aqueous solutions of electrolytes is a long-standing problem of electrochemistry. This problem is more difficult than the case of liquid metals or molten salts where the number of distinct atom or ionic species is n < 3. For a molten salt such as KC1, with c

s


17.

BAIANU ET AL.


EXCHANGE RATE Intermediate

Fast


Slow

150

0

-150 150

0

-150 150

0

-150

FREQUENCY (kHz) Figure 6. Simulated deuterium quadrupole-echo lineshapes under various exchange conditions. In all cases the spacing between 2 με 90° pulses is 50 μς, and all spectra have been normalized to unit intensity. The static quadrupolecoupling parameters are e qQ/h =180 kHz and η = 0.06, as expected for orthoor meta- deuterons on a benzene ring. The motion in model A is simply 180° flips about the Q-C4 axis of the ring, while models Β through Ε all include additional motion as described in the text. "Slow", "intermediate", and "fast" refer to flip rate, which was stepped from 103 to 105 to 107 s" for each model. A set of experimental lineshapes which exhibited features of models B,C,D, or Ε could not be adequately reproduced by the two-site model A. (Reproduced with permission from ref. 10. Copyright 1987, Academic Press). 2

1


279

MOLECULAR M O D E L I N G

280

n = 2 , the number of pair correlation functions that characterize completely the local s

structure is 3, whereas for molten LiOH, n = 3 and the number of atom-pair correlations required is 6 (Figure 7a); four additional correlation functions and 2Dcorrelation functions were needed to define the coordination of Li and OH" in molten LiOH, and to represent the complex medium range structure (r -10 Â) caused s

c

by the anisotropic (directional) nature of the OH~ ion. For an aqueous solution of electrolytes ten atom-pair correlations are needed. Since X-rays are scattered only


1

2

very weakly by H and H atoms, the atom-pair correlation functions involving these atoms are unobtainable from X-ray diffraction experiments. However, the atom-pair correlation function, goo(r) for the oxygen atoms can be reliably obtained from X-ray diffraction measurements for concentrated solutions. To determine the remaining pair correlation functions that involve H (or H) one needs to carry out neutron scattering measurements also. Few high quality/high Q, neutron diffractometers were built because of the very high cost. Both neutron scattering measurements and the analysis/corrections of neutron scattering data are relatively complex, timeconsuming and require careful scrutiny (/ /). As a result, progress has been relatively slow in solving this problem, and a larger number of X-ray scattering than neutron scattering studies were carried out. An important development in the neutron scattering study of electrolyte solutions was the selective use of isotopic substitution to obtain the partial, pair-atom correlation functions (11). Among the most studied systems are the aqueous solutions of LiCl and those of N1CI2. Figure 8a shows the X-ray scattering intensity as a function of the modulus of the scattering vector, s = 4π8ΐηθ/λ (Â ), for two aqueous solutions of concentrated N1CI2. Note the limitation to s-values of less than 10 Â in this data set and the limited resolution in the corresponding total correlation function, G(r), (Figure 8b). The X-ray scattering curve for a metallic glass (12) (FeNiPB, Metglass 2826) with the same number of atom species (n = 4) is shown in Figure 8c for comparison with Figure 8a. Note the apparently simpler X-ray scattering function for the FeNiPB metallic glass and the small peak at low angles, near 1 Â" in Figure 8c. Annealing of this glass causes slow structural relaxation (broadly similar to the case of molten LiOH discussed above) which diminishes the intensity of the 1 Â" peak and sharpens up the rest of the pattern. The limit of s in such measurements was extended to 14 A" (moderate resolution); an X-ray scattering curve to a higher svalue of 17.4 Â" (high resolution) is shown in Figure 8d for a metallic glass of simpler composition, C09P, (13) (n = 2). Both sets of data in Figures 8c and 8d were obtained by an energy-dispersive technique so that the r.d.Fs can be directly determined without uncertain corrections of the scattered X-ray intensity for Compton scattering (14). Neutron scattering curves are shown in Figure 9a for a series of concentrated electrolyte solutions in water. The N1CI2 scattering curve has two lower peaks near 3 À" , following the main peak (instead of one peak as in the case of the other solutions). From such data, but at higher resolution (Q « 16 Â" ), the neutron partial structure factor SNÎNÎ was derived for aqueous solutions of NiCl at various concentrations (Figure 9b). The effect of isotopic substitutions on the neutron scattering intensity at the main scattering peak near 2 Â" is shown in Figure 9c, both for the anion (CI") and the -1

s

1

1

1

1

s

1

1

1


17.

BAIANU ET AL.


281


a 3

Figure 7. (a) Atom-atom pair correlation functions for molten LiOH at 767 K. (b) Calculated (solid line) and experimental (broken line) neutron weighted structure functions Q[S(Q)-1] for molten LiOD. (Reproduced with permission from ref. 30. Copyright 1990, the American Institute of Physics).

I, -80

/

\

~

65

-70 5 5

\

" -60

\V

-50

S ft -40

y1 \\ \ \ /

\

~ Ο

\

25-30

-20 \ !

1 _ !

ι

2

3

5-

1

1

1

l\ !

!

I

A

5

6

7

9

10

θ

Figure 8a. Intensity curves (e.u.) for two N1CI2 solutions in water. Top curve is 3 3 for 4 mol dm" , and bottom curve is for 2 mol dm" . (Reproduced with permission from ref. 6. Copyright 1977, the Royal Society of Chemistry: Cambridge, UK). In Molecular Modeling; Kumosinski, T., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 1994.

MOLECULAR MODELING


282

Figure 8b. Radial distribution function derived from the data in Figure 8a for 2 aqueous solutions of N1CI2. (Reproduced with permission from ref. 6. Copyright 1977, the Royal Society of Chemistry: Cambridge, UK).



BAIANU ET AL.


4

8 10 12 11

β

SdnaieMF)—

Figure 8c. X-ray intensity functions of Metglas 2826, before and after annealing at 628 Κ for 30 min (intensity corrections as described in the text); the intensity oscillations between 13 and 17 À" were within the noise level; 1

1

intensity measurements between 17 and 22 À" were made by a resonance method - to be published (Agk ^ rotating anode / Ru fluorescence filter) a

showed again a smooth decrease, without marked oscillations as those seen in intensity functions derived from EDXS measurements); — as received, — annealed.

0

1

2

3

4

5

6

7

9

10 11

12 13

14

15 16

-1 Figure 8d. The X-ray interference function of noncrystalline C00.9P0.1 at 293 K, obtained by energy-dispersive techniques. In Molecular Modeling; Kumosinski, T., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 1994.

283

284

MOLECULAR MODELING


f\

NiCI

2

BQCI

2

KCI

./ Λ ,

L

;

0

NaCl

'i

LiCI

Λ /

_l

^

1

I

2

I

3

I

A

'

5

•

6

I

7

8

I

9 I

L_

Figure 9a. Intensity (in arbitrary units) versus scattering angle of thermal neutrons for a variety of solutes in heavy water. All the solutions are close to saturation. The patterns are similar because the bulk of the scattering arises from water. (Reproduced with permission from ref. 11. Copyright 1978, IOT Publishing: Bristol, UK).


17.

BAIANU ET AL.


285

90! 60 tltiliiintii

30| 0

"

H

-

4 J M V.

^J'IIIUHMHIUI

m m >

m

m

~ u


-30

3.5 ^^rnirTtii^^'^tirnmimiiiiiiiiiiiiiir

3.1

90 60 30

o| -30

θ

10

12

14

16

Figure 9b. The variation with Q of the partial structure factor SNÎNÎ for aqueous solutions of N1CI2 of various percentage compositions. (Reproduced with permission from ref. 48. Copyright 1973, Elsevier Science Publishers B.V.: Amsterdam, The Netherlands).


MOLECULAR MODELING


286


17.


BAIANU ET AL.

287

2+

cation (Ni ). This difference in the neutron scattering intensity for different isotopes gives the basis for the 'difference method' of local structure determination of electrolyte solutions by neutron scattering combined with isotopic enrichment. For a salt M X dissolved in D2O, with M being the cation and X- being the anion, one can calculate the effect of isotopic substitution of the cation on the neutron scattering intensity in absolute units. The difference in neutron scattering cross-sections between two identical solutions but with different isotopes of the cation, Μ , can be written as (15): +

n

A ( Q ) = AM(Q) + CORRECTION TERMS

(3)


M

where the correction terms are inelastic and incoherent scattering terms, and AM(Q) is the difference between the coherent neutron scattering intensity components of the two D2O solutions made with different isotopes of M. The function AM(Q) is called the first-order difference and can be calculated in terms of the atomic fraction c of M, appropriate (coherent) scattering factors, bM, bM, bo, bo, and partial structure factors Saji(Q): *M(Q) = Ai[S o(Q)-l] M

+

B![SMD(Q)-1]

+

CitSxMÎQMl+D^SMMÎQ)-!]

W

where Ai, Bi, Q , and Di are expressed in terms of c, η and the appropriate products 12

of coherent scattering amplitudes, b (in 10" cm units). Thus, Ai = (2/3)c (1 - c - η · c) · b (b - bjti), for the S o term, 0

Bi

=

M

(5)

M

4

( /3) · c(l - c - η · c)bo · 0=>M - b|vi), for the SMD term,

(6)

2

Ci = 2nc · bx(bM - b^), for the S M term,

(7)

X

and 2

2

Di = c · [(bju - (b^,) ], for the S

M M

term,

(8)

For the isotopic substitution of the anion X~, a first-order difference, Ax(Q), can be written in the same form as the above equations, with the appropriate (bx-bxi) difference appearing in the products of the coherent scattering amplitudes, instead of 37

35

ΦΜ-^Μ)· As an example, the isotopic substitution of C1 with C1 would give a 12

62

difference (b -bx) of (1.18-0.26) = 0.92 χ 10" cm, whereas the substitution of N i x

with natural Ni (mixture) would provide a difference (bM-bM) of [1.03 - (-0.87)] = -12

1.90 χ 10 cm. Note that the method is twice as sensitive to the nickel cation substitution than to the CI anion substitution. The use of the natural CI isotope mixture and C1 yields a difference (b -bx) of only 0.70 χ 10" cm. 37

12

x


288

MOLECULAR MODELING

In order to derive the Sxx(Q) and SMM(O) to test for ordering of the ions in solution one would need to employ three different isotopes for either Sxx or SMM determination. The cross-term SMX(Q) requires four different isotopes. Furthermore, a second-order difference method is required to determine Sxx, SMM

O R

SMX-

The ionic hydration is best understood in real, rather than scattering space; therefore, the Fourier transforms (FT) of AM(Q) and Ax(Q) are required to obtain the +

corresponding distribution function, G. For the cation, M : 2

G (r) - l/27T g(r). JA' (Q) · Q sin(Q-r) · dr Downloaded by MONASH UNIV on October 26, 2012 | http://pubs.acs.org Publication Date: December 14, 1994 | doi: 10.1021/bk-1994-0576.ch017

M

(9)

M

or G (r) - A!(gMD(r) - D + Di (g (r) - D M

(10)

MD

Because Ai and B\ are much greater than Q and Dj one has that G M W - Ai (g (r) - 1) + Bi (g (r) - 1) MO

(11)

MD

+

which would allow one to locate the water nuclei (O and D) around the cation M . Examples for N i and L i are shown, respectively, in Figures 10a, 10b and 10c for concentrated N1CI2 and LiCl solutions in D2O (16,17). A similar determination was carried out for the Cl~ anion in NaCl and CaCl solutions in D 0 (17); the weighted distribution G (r) - A (gcio(r) -1) + B (gciDO0 -1) C (gaci(r) -1) + D (gaci(r) 1) is shown in Figure lib for a 4.5 mol solution of CaCl in D 0. Related X-ray scattering data and calculations are shown in Figures 11a. The first-order difference, ^Ni(Q), from which the weighted distribution GNÎOO was obtained by FT (Figure 10a), is shown in Figure 12; this difference is well-behaved at high Q ( 18). The main features of the weighted distribution function GNÎW (Figure 10a) are the Ni-O peak near 2.1 À, the Ni-D peak near 2.7 À and the second nearest-neighbor peaks near 4.5 Â . The water coordination in thefirsthydration shell of N i is shown 2+

+

2

2

+

a

2

2

2

2

2

2

2+

above the GNi(r) curve and involves six water molecules that have a tilted DOD bisector axis by an angle φ = 34 ± 8° to the Ni-0 axis. In solutions of concentrations lower than 1 molal the Ni-D distance increased suggesting a distortion of the 2+

hydration shell of Ni as the packing fraction of hydrated ions is increased ( 16). This observation may help with the selection of more realistic choices of the interionic potential than those currently employed in MD or MC simulations. The coordination of water around the chloride anion (Figure lib) also involves a tilting of the water molecule OD axis from the Cl-O axis by an angle ψ of 5° or less ( 15, 16). The Cl-O nearest-neighbor peak in Figure lib had a maximum at 3.20 ± 0.04 Â in agreement with the X-ray scattering data for CaCl2 solutions in H2O (18). For NiCl solutions the CI" was apparently absent from the first coordination shell of Ni . A second 2

2+

hydration shell around N i was also suggested by both neutron diffraction and quasielastic neutron scattering studies (15, 20, 21).



17.

289


BAIANU ET AL.

1 I

2

3

I

I

4

5

I

6

I

7

ι

8

I

L

Figure 10a. The weighted distribution function G i(r) = Al(gNio - 1) + B l ( g N - , - 1) + Cl(gNia - 1) + D l ( g N i i - 1) for a 4.41 molal solution of NiCl obtained by taking the Fourier transform of the data in figure 9a. At this concentration the coefficients A l · · *D1 are respectively 17.4, 40.0, 5.05 and 0.32 mol. The conformation of N1-D2O consistent with this ΘΝΪ(Γ) is shown at the top right. (Reproduced with permission from ref. 11. Copyright 1978, IOT Publishing: Bristol, UK). N

0

2

N

0-030,

0*030l

1 VO

1 2-0

1 3-0

1 VO

1 5-0

1 6-0

1 >0

I

Figure 10b. G (r) for a 3.57 molal solution of LiCl in D 0. Top right: Li

2

the L1-D2O conformation consistent with Gy(r) shown infigures1 and 2 was that with the φ values given in table 2 of ref. 15. (Reproduced with permission from ref. 15. Copyright 1980, the Institute of Physics: London, UK). In Molecular Modeling; Kumosinski, T., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 1994.

MOLECULAR MODELING

290 0Ό4

-

0-02 Γ

α

η

\ f^- '

~% 0-00 t.

ri? :


!: Ί· V ';: -0-0 2

h

ι

-0-CU

1·0

ι 2·0

ι 3·0

ι

1 5·0

4·0

_l

1 7-0

6-0

+

Figure 10c. Hydration structure of L i and Cl~ ions in concentrated solutions in D 0 derived from neutron scattering isotope difference measurements. (Modified from ref. 11). 2

The analysis of the first-order difference results (7/) for concentrated LiCl +

solutions in D 0 showed that the orientation of D 0 around CV and L i (Figure 10b) 2

2

is similar to that of D 0 in NiCl solutions (Figure 10a). The hydration number for 2

2

+

L i was strongly concentration dependent and increased to 5.5 below 3.6 mol (Table I); these neutron diffraction results did not seem to support tetrahedral coordination of D 0 around L i . Hie possibility still exists that a hydration of 4+2' is involved, with +

k

2

a first hydration shell which is tetrahedral and a second hydration "shell" of two D 0 molecules differently oriented in comparison with either the water molecules in the bulk or in the first hydration shell of L i . It was also found that water-bridging between L i and CI" ions occurs at the higher concentrations of LiCl, in agreement 2

+

+

with the results of previous work by pulsed *H NMR on LiCl · η Η 0 glasses at 100 Κ (3). Thefirst-differenceneutron diffraction results for aqueous solutions of concentrated electrolytes disagree with the results of MD simulations of electrolyte solutions in water which employed the ST2 water potential (77); the latter favor a φtilt angle of water around cations of 55°. The results from other MD calculations were also compared with the local structure of LiCl solutions in D 0 derived from the neutron diffraction data with the first-order difference; partial agreement with some of the MD simulations was found (30), (Table Π). Previous X-ray and neutron scattering results for aqueous solutions of LiCl (75) arrived at somewhat different conclusions 2

2


17.

BAIANU ET AL.


291


G(r)

Figure 11a. Radial distribution functions for CaC^, derived from X-ray scattering studies of concentrated (1M, 2M and 4M) CaCl studies in water. (Reproduced with permission from ref. 18. Copyright 1976, the American Institute of Physics). 2

based on the total distribution functions, g(r), (Figures 13a and 13b, respectively). At high Li CI concentrations (η £ 16) it was shown (43) that the nearest-neighbor tetrahedral coordination of hydrogen-bonded water molecules was absent because there was no peak maximum near 2.8 Â in the g(r) derived from the X-ray scattering measurements. The second-neighbor peak for tetrahedrally Η-bonded water near 4.5 Â was also absent for η £ 16 in LiCl*nH20 solutions. Η-bonded, tetrahedral water species were present, however, in the more dilute LiCl · η Η 2 θ solutions, with η > 33, suggesting the presence of two different environments for water at these LiCl concentrations; the first type of water coordination was that characteristic of the Li and Cl~ (separate) hydration shells, whereas the second type was characteristic of bulk, liquid water.



292

MOLECULAR MODELING

Γ (A) I

?

1

2

4

1

I

6

fi

ι

Ï

Figure lib. The weighted distribution Gci(r) = A2(gcio - 1) + B2(g - 1) + C2(gaa - 1) + D2(gcici - 1) for a 4.49 mol solution of CaCl . At this concentration the coefficients A2 · · *D2 are respectively 16.0, 36.8, 1.16 and 3.77 mb. The conformation of CI-D 0 consistent with this Gci(r) is shown at the topright.(Reproduced with permission from ref. 11. Copyright 1978, IOT Publishing: Bristol, UK). aD

2

2


17.

BAIANU ET AL.

293


ΚΜ)5

•v Downloaded by MONASH UNIV on October 26, 2012 | http://pubs.acs.org Publication Date: December 14, 1994 | doi: 10.1021/bk-1994-0576.ch017

U

\

-0.05

r-o.10

12

10

H

16

Figure 12. The first order difference for two 4.41 model NiCl solutions made 2

62

nat

from N i and Ni (native mixture). It should be noted that this difference is well behaved at high value of k. (Reproduced with permission from réf. 11. Copyright 1980, IOT Publishing: Bristol, UK). Table L Hydration of L i Mol Lithium-Oxygen Distance

+

Hydration Number

Lithium-Deuterium Distance

9.95

1.95 ± 0.02; 2.1 ±0.05

2.50 ± 0.02; 2.6 ± 0.05

52° ± 5 ; 35° ± 5°

3.3 ± 0.5; 4 ± υ.5

3.57

1.95 ± 0.02

2.55 ± 0.02

40° ± 5"

5.5 ± 0.3

ΰ

a

φ is the angle between the plane of the water molecule and the Li-O axis. It was assumed that TOD is 1 Â and DOD was 105.5° in ref. 3, whereas DOD was assumed 109.5° in ref. 3 and 5, and Too was determined to be 0.95 Â . SOURCE: Reprinted with permission from ref. 15. Copyright 1984.


294

MOLECULAR MODELING

An independent-hydration model was proposed for the hydration of ions and for the analysis of the diffraction data for LiCl · 11H2C). The geometry of water coordination in this model is as specified in Figure 14. Additional results obtained with this modelfromX-ray scattering data for LiCl, liBr, CaCl and CaBr solutions (19) are summarized in Table ΠΙ. The neutron diffraction data for l i C l n D 0 solutions (21) were analyzed with the same model (19) and the corresponding g(r)'s are shown in Figure 13b. The first OD peak has a maximum near r = 1 À in Figure 13b; the comparison of Figures 13a and 13b provides a rough estimate of the Cl-O and Cl-Cl pair-correlation 2

7

2

e

2

7

contributions to g(r) of IiCl · nD 0 for η ^ 16. The peak near r = 3.2 Â in the g(r) of 2

7

molten LiOD is broader and of lower amplitude than the corresponding peak for Downloaded by MONASH UNIV on October 26, 2012 | http://pubs.acs.org Publication Date: December 14, 1994 | doi: 10.1021/bk-1994-0576.ch017

7

7

IiCl-3.0D O or IiCl-4.1D 0 in Figure 13b. 2

2

The tetrahedrally coordinated

+

oxygen atoms around I i are, therefore, only partially responsible for this peak in the 7

g(r) of LiCl-nD 0. The Cl-O pair correlation is expected to contribute also to the 2

7

3.2 Â, asymmetric peak in the g(r) of LiCl solutions in D2O; note also that the 3.2 Â 7

peak in the g(r) of molten LiOD is, on the other hand, symmetric. The 3.2 Â distance for the Cl-O pair correlation nearest neighbor peak would be consistent with the randomly distorted octahedral coordination of six water molecules packed around the CI", as suggested schematically in Figure 14c.

The "independent-hydration" +

model, however, neglected the water-bridging between L i and CI" which becomes important in concentrated (n £ 16) LiCl*nH 0 (or D 0) solutions; the correlation radius, derivedfromFigures 13a and 13b extends to about 7 Â in the concentrated solutions and is about the same as in liquid water (or D 0) at 293 K. 2

2

2

Molecular Dynamics of Aqueous Solutions of Electrolytes.

Monte Carlo

Simulation of Ion-Water Clusters. The isolated ion-water clusters are relatively simple models for aqueous solutions of electrolytes. Their study is likely to provide potential functions and parameters for the interactions of ions such as l i , Na , K , +

-

-

-

-

9+

+

+

F , Cl , Br , I and Ca with water, which could be of use also in MD or MC simulations of aqueous solutions of such electrolytes. The isolated ion-water clusters for alkali and halide ions were intensively studied experimentally by mass spectrometry (22-23) which provided estimates of the Gibbs free-energies, AG°(N-1,N) for the formation of ion-water complexes, as well as enthalpies of reaction for the ionic hydration. There have been several attempts to calculate these quantities by employing empirical potential function techniques and quantum mechanical methods, most of which were based on two-body potential functions. The neglect of the many-body effects causes the calculated values of the interaction energy to be too negative in comparison with the experimental values, and this effect increases with cluster size or ionicfieldstrength. Recently, non-additive potentials were employed together with methods of statistical mechanics that involve MC computations of the free energy perturbation in terms of a series of moments of the potential energy function (24). The total


17.

BAIANU ET AL.


295

Table Π. Lithium-Water Configurations in LiCl Solutions Mol or Hydration Method Ion-Water Ratio r io (Â) Number ruH (Â) L

2.2 mol

2.10

2.60

7

Molecular dynamics (ST2)

2:50

2.20

3.30

5

Cluster calculation

2:200

2.00

2.60

5

Cluster calculation

Single L i ion

1.81-1.89

—

—

Ab initio

1:6

—

—

4.0

Electrostatic Model

1:4

—

—

4.0

Experimental, gas phase

1:4

2.1 ±0.05

2.6 ± 0.05

4 ±0.5

X-ray/Neutron scattering

1:2 to 1:12

2.1 ±0.05

2.6 ± 0.05

4 ±0.5

Pulsed H NMR


+

l

Table HL Mean Distance, rcs> and Mean Square Deviations, ocs> for Cation-Solvent Interactions in the Groups Cation (H 0) , and Mean 2

n

Distances, TAS> and Mean Square Deviations, Ό\$, for Anion-Solvent Interactions in the Groups Anion ( H 2 0 ) Solutions

m

a

η

m

res

ocs

TAS

LiCl -13.90 H 0

4

6

1.99

0.28

3.04

0.17

LiCl-8.15 H 0

4

6

1.95

0.25

3.10

0.20

LiCl-6.44 H 0

4

6

2.04

0.04

3.09

0.22

L i Q - 4.01 H 0

4

6

2.22

0.31

3.18

0.19

Lia-24.97 H 0

4

6

2.25

0.25

3.29

0.23

LiBr-24.97 H 0

4

6

2.14

0.25

3.29

0.25

LiBr-10.83 H 0

4

6

2.16

0.25

3.29

0.26

LiBr-8.41 H 0

4

6

2.16

0.25

3.29

0.26

CaBr -44.05 H 0

6

6

2.40

0.12

3.32

0.24

CaBr -25.95 H 0

6

6

2.44

0.15

3.34

0.26

CaCl - 55.82 H 0

6

6

2.42

0.14

3.14

0.22

CaCl -26.62 H 0

6

6

2.41

0.15

3.14

0.23

CaCl -12.28 H Q

6

6

2.42

0.13

3.15

0.21

2

2

2

2

2

2

2

2

2

2

2

2

2

2

2

2

2

2

a

Obtained from least-square refinement. SOURCE: Reprinted with permission from ref. 18.


296

MOLECULAR MODELING

H 0 2

LJÇ1 > 136

H2O

1101*68,0 H 0 2

L i C l «33.3 H Q

χ δ

2

L i C l * 16.2 H Q 2

LiCl'8.15H Q 2

LiCl'4.01 H Q 2

UC1*3.00H 0


2

1

2

3

4

5

6

7

8

9

10

Γ/λ (α) Figure 13. The radial distribution functions g(r) measured with: (a) X-rays and (b) neutrons for aqueous solutions of lithium chloride of various concentrations. The individual curves are displaced vertically to avoid overlap. (Reproduced with permission from ref. 21. Copyright 1973, the American Institute of Physics).

CI (H 0) 2

6

CO U*(H 0) 2

4

Figure 14. Diagram (a) shows the environment of water molecules in pure water. Diagrams (b) and (c) show, respectively, the four-fold and six-fold coordination by water molecules of the L i and CI" ions in concentrated solutions (c). (Reproduced with permission from ref. 21. Copyright 1973, the American Institute of Physics). +


17.

BAIANU ET AL.

297


interaction energy for the ion-water cluster was expressed (24) as: Etol

=

Ewater-water pair

+

Epater-ion pair

+

F -j _ i - t r ex

0n

wa

er

wa

e

+

Epol

(12)

where E _ pair is a RWK2 pairwise additive potential, modified to include nonadditive terms. Such a model yields reasonable values for the second virial coefficient and water-dimer energy in the gas phase, as well as lattice densities and energies for ice L, and ice VII. The polarization term was (44): w a t c r

w a t e r


Epo^d/ÊajtEj.Ej]

(13)

where aj are the polarizabilities of the polarizable centers within the clusters, and Ej is the electric field created at a center j by the surrounding molecules. MC simulations were carried out at 298 Κ for Na , K , Mg , F", and CI" ions hydrated with a variable number of water molecules between 1 and 6. The best agreement with experimental values of the enthalpy was obtained for the larger ion, K , the least +

+

2+

4

+

satisfactory agreement was for L i , although the results were improved over those obtained previously with two-body potentials. The structural features of the ion-water clusters obtained with the nonadditive potentials are summarized in Table Π. The first hydration shell of Li was tetrahedral (n = 4.0) and there were 2 additional water H

+

+

molecules outside thefirsthydration shell of L i . The Na had a well defined *5+Γ +

structure, whereas the K hexahydrate complex was between the "4+2* (like +

+

Li (H 0) ) and the *5+Γ structure (like Na (H 0) ). Fluoride hydration involved 4 2

6

2

6

water molecules in a single hydration shell, whereas Cl~ had a well defined structure fc

of the type 3+l\ The water coordination around F" or Cl~ was not, however, tetrahedral but there were instead 'favorable* water-water interactions present only arounçl the anions. It was suggested that in electrolyte solutions the coordination number would be higher than for the isolated ion-water clusters, because in solution the addition of water molecules to the first coordination sphere of ions would be energetically compensated more effectively by polarizing water molecules that are located in the second hydration shell (24). The computation method employed for isolated ion-water clusters could also be applied to study ionic solutions, and it will be interesting to see such results, especially since at the moment the only extensive MD reported for ionic solutions are with the MCY or earlier potentials (25, 26).

Orientational Correlation Functions for Ionic Solutions in Water (42). The second-rank orientational correlation function cf(t), (/* = 2 in reference (/)) were also calculated in MD simulations with the MCY potential (2) for hydrated ions and showed similar time dependences (Figures 7 and 8) in reference (2)) with that found in liquid water (Figure 1); the exponential decay for t > 0.1 ps was, however, slower for all the ions investigated ([Li ] , [F~] and [Cf] ) in comparison with liquid +

aq

aq

aq

H 0, and the τ values increased with decreasing ionic radius (increasing ionic field 2

Γ

strength), as expected. Calculated reorientation times τ for the various hydrated ions (26) are summarized in Table IV corresponding to the a (t), (a = x, y or z) orientational correlation functions for ionic solutions in water around 0.8 M. Note Γ

2


298

MOLECULAR MODELING

that the value of

(or

for liquid H 0 (2.01 ps) at 286 Κ is smaller by about a 2

factor of 2 than the experimental value of T of water determined by NMR relaxation measurements at 286 K. The dipolar correlation functions were also calculated for ionic solutions in water and the values of the dipolar correlation time ("relaxation time"), τ*, are summarized in Table V. Note that τ | > i\ for liquid H 0 at 286 K; c

2

however, the calculated

1.9, which is less than the expected value of 3

TI/TJ=

expected for this ratio from the experimental

Tdiek-etric

/T^

M R

value, as well as

from dielectric relaxation theories (27, 28). Both values of calculated i\ and

are

too short in comparison with Tdielectric and T ^ . This may be caused by the fact that collective effects are ignored in the MCY potential, which are however, important in the OCF's involved in dipole reorientations and dielectric relaxation in water. Downloaded by MONASH UNIV on October 26, 2012 | http://pubs.acs.org Publication Date: December 14, 1994 | doi: 10.1021/bk-1994-0576.ch017

M R

a

Table IV. Calculated f-2 Reorientation Tiraes

Time (ps) Ion Li

Temperature (K)

+

Χ

Y

Ζ

3.7

5.8

8.0

h

297

Α τ

2

298

Α τ

2

1.4

2.3

2.1

274

A r

2

1.1

2.1

1.4

F"

278

Α τ

2

3.9

4.8

5.2

cr

287

Α τ

2

1.1

2.1

1.4

H0

286

Α τ

2

1.03

2.01

1.07

Na K

+

+

2

2

2

2

2

2

2

a

Defined as the integral of the normalized correlation function.

The set labeled Y is related to the NMR intramolecular dipole-dipole relaxation time. SOURCE: Reprinted with permission from ref. 26. Copyright 1982. Table V. Reorientational Correlation Times (in ps) Run

TÏ

TÏ

τϊ

τ

χ

Τ2

2

TNMR

1

9.0

16.3

10.3

4.3

7.0 (5.7)

4.4

2

3.2

6.0

3.5

1.8

2.7(2.1)

1.7

3.6

3

3.6

6.3

3.6

2.0

2.8 (2.2)

1.8

4.8

4

3.1

5.7

3.8

1.7

2.7(2.0)

1.8

5

1.0

1.9

1.2

0.6

0.9 (0.6)

0.6

a

b

0.9°

Bracketed numbers are results for A^ τJ. a

H 0 at 283 Κ; *ΐ> 0 at 283 Κ; Ή 0 at 363 K. SOURCE: Reprinted with permission from ref. 26. Copyright 1982. 2

2

2


17.

BAIANU ET AL.

299


Residence Times of Water in the Hydration Shell of Ions. The residence time, T R , is defined as the mean time for which a water molecule remains in the first hydration (or coordination) shell of an ion before exchanging with the bulk water population. In the absence of a bulk water population, as in the case of very N

concentrated solutions, T R would be determined by the mean time in which a specified water molecule diffuses outside the nearest-neighbors (coordination) peak in the ion-water oxygen pair correlation function, gion-oxygen, discussed in the previous section. For example, a residence time of about 30 ps was estimated for water in the hydration shell of I i based on NMR relaxation measurements (28). The MD simulations can also estimate the residence time either from the van't Hove (distinct) space-time correlation, Gôn-oxygenO*, t), or by defining (23) a function Qj(t, t ; t*) for a water molecule j; the latter function takes the value 1.00 if the molecule j is within the first peak of gion-oxygen(r)> (the first hydration shell of the ion) at both timesteps t and (t+tn), and it is zero otherwise. If the water molecule resides outside the hydration


n

n

shell for only very short times t* « t then Qj(t, t ; t*) = 1.00. A suitable value for t* would be around 2 ps, the time interval in which water molecules in the bulk exchange neighbors, (t* is a sort of " T R " in bulk water). The exchange of the water molecules with the bulk water is then characterized by: m

n

Nt îlion(t) = 1/Nt) Σ Σ Qj(tn, t; t*) n-1 j

(14)

where there are η time steps in the MD simulation (25). At t = 0, nion(0) is the average coordination (or hydration) number of the ion; nionOO would give the mean number of water molecules that are left within the hydration shell at the time t\ For both alkali and halide ions, the function ni (t) decays exponentially at short times: on

nion(t) = n (0) · exp(-t/TR)

(15)

ion

where nion(0) = n is the average hydration number of the ion. Calculated values of the residence times for alkali halide solutions in water are summarized in Table VI, and were found to be in quite good agreement with values estimated from NMR relaxation measurements. H

Velocity Correlation Functions of Ions in Aqueous Solutions. The VCF's of water in various solutions of alkali halides were calculated for water in the first hydration shell of the ions (26). Their time dependence is not much different from that of liquid water. On the other hand, the VCF's of the ions in aqueous solutions (Figure 15) are quite different from the VCF of water (Figure 9a and 9b in reference 1). Upon FT of the VCF's of [Li ]aq and [Na ] shown in Figure 15 new +

+

aq

1

+

1

+

frequencies appear near 400 cm" for [Li ] , near 260 cm" for [Na ] (and [F~]aq). The peak frequency seems to be correlated with the reciprocal of the ionic radius. Raman studies by Nash et al. (29) for solutions containing [Ii ]aq reported a band aq

aq

+


300

MOLECULAR MODELING

1

+

near 380 cm" which was assigned to the F mode of the L i ion "in its tetrahedral 2

1

cage" (of water molecules), and a band near 440 cm" assigned to the "Al mode of the coordination complex", (the so called 'breathing' vibration of coordinated water). Somewhat similar results were obtained in very concentrated LiCl · nH 0 solutions (2 < η < 12) at 298 K. If these assignments (29) were correct, than the "power" spectra calculated by FT of VCF's for the hydrated ions would be in reasonable agreement, in terms of the peak position for [Li ] . Such results are discussed next in more detail. 2

+

aq

Table VI. Residence Times Calculated for Monovalent Ions and Water.


Ion Li

+

Li

+

Na K

T (ps)

T(K)

R

33 6.0 10 4.8 20 4.5 4.5

278 368 282 274 278 287 286

+

+

F"

cr H0 2

Raman Spectra of L i C l - n H 0 Solutions and Glasses. Since the MD simulations with the MCY potential for ionic solutions in water (31) did not consider the intramolecular vibrational modes of water molecules (water molecules were 'rigid'), the important OH-stretching region could not be compared with the corresponding bands in the Raman spectra of ionic solutions in water. Such a comparison between the theory and experiment was, however, made for molten alkali halides, and could also be carried out for the vibrational correlation function (26) C (t), as well as for the rotational autocorrelation function (34), C (t). In this context, it will suffice to consider the main characteristics of the Raman spectra of LiCl«nH 0 solutions and glasses; a quantitative analysis of such spectra will be reported elsewhere. The OH-stretching region of the polarized (VV) Raman spectrum of a Li0-4.1H 0 solution at 293 Κ is presented in Figure 16a; the corresponding region of the VH Raman spectrum is shown in Figure 16b. The OH-stretching (VV) bands of liquid water at 293 Κ are shown for comparison in Figure 16c. The liquid water 2

v

T

2

2

has a distinct band near 3200 cm" , whereas this band is nearly absent in the LiCl»4.1H 0 solution Raman spectrum. Since the OH-stretching band near 3200 2

-1

cm was assigned to the tetrahedrally Η-bonded water (HB) species, this implies that the tetrahedraly Η-bonded water structure is 'broken', or absent, in the LiCl-4.1H 0 solution, in full agreement with the r.d.f's derived from X-ray and neutron scattering studies of concentrated LiCl solutions in water as (D 0), (Figure 13a and 13b). It is 2

2


17.

BAIANU ET AL.


301

rather mteresting that, upon cooling the concentrated LiCl*nH20 (2.2 £ η £ 11) solutions to 100 K, these form a glass, without phase separation/ice formation (5). The OH-stretching region of the Raman sr^trum of a LiCl · 4H2O solution at 293 Κ (Figure 16a) is compared with that of the coiresponding glass at 119 Κ (Figure 16d), close to the 'rigid lattice' condition (in which any large amplitude motions are frozen out). Clearly, several OH-stretching bands increase in intensity at 119 Κ relative to the non-tetrahedral ("NHB") water bands near 3450 cm" and 3400 cm in Figures 16d and 16e. (Their precise positions will require a Gaussian deconvolution of the Raman spectrum after appropriate intensity corrections.) Unlike the Raman spectrum 1

_1

of glassy water at 90 Κ (reference 1), the W spectrum of glassy LiCl · 4H2O at 119 Κ 1


(Figure 16d) lacks the 3080 cm" band assigned to the vf* symmetric OH-stretch of tetrahedrally Η-bonded water. This confirms the absence of ice Ih, cubic ice (Π/ΠΙ) and 'glassy' water in the IiCl-4H 0 glass at 119 K. The intense band near 3200 2

cm

-1

which was present in supercooled, liquid water at -33°C (240 K) is also absent

in the Raman spectrum of LiCl · 4H2O glass at 119 K, or it has lower amplitude than the other bands. This suggests very strongly that the local structure of the LiCl · 11H2O glasses is quite differentfromthat of various ices, 'glassy' water, supercooled liquid water, or liquid water. The absence of tetrahedrally, Η-bonded water in the LiCl-nH20 glasses is, therefore, strongly suggested by these Raman scattering observations.

Furthermore, significant differences are observed between the 1

IiCl-nH20 solutions and glasses in the OH-stretching region from 3250 cm" to 1

3450 cm" in the polarized Raman spectra. Such differences seem to suggest that the bb

1

'bent-bonds', responsible for the v , OH symmetric stretch (3370 cm" band), occur morefrequentlyin the LiCl · 4H2O glass at 119 Κ than in the corresponding solution. Both the IiCl*nH20 solutions and glasses have in common with liquid water, supercooled water, 'glassy' water and the various ices, the OH-stretch band near 3,450 cm" assigned to the v^ asymmetric stretch mode, with one strong and one weak bond. However, the 3,450 cm" band had much lower height in 'glassy' water at 90 Κ than the tetrahedrally, Η-bonded water band near 3080 cm" , whereas in 1

b

1

1

LiCl ·ηΗ2θ solutions and glasses the 3,450 cm" band is the tallest one and seems to dominate the spectrum. This suggests that the dominating water species in the LiCl · Î1H2O solutions or glasses have the following bond scheme: 1

H Q ' / : Li · · · Ο strong «-*weak \ : Η Η

\

Ο—Η

H Cl" / I i · · · Ο 3), when the solubility limit permitted (Figure 24a). Note that these concentration dependences were different from those observed for the LiCl · RD 0 or CsF, KF, RbF. RD 0 solutions (Figures 23a and 24c, respectively). It would seem that the large sizes of the perchlorate and tetrafluoroborate anions prevent, in these cases, the association with the cation into well-defined, hydrated η-mers; however, at high concentrations water molecules should still be bridged between the counterions (Figure 25). Therefore, MD simulations have not yet been reported for these complex anions in aqueous solutions. 2

2

Molecular Dynamics and Nuclear Spin Relaxation Studies of Glycine Hydration and Activity in Aqueous Solutions. This is the first report of extensive 17 2 Ο and H NMR transverse relaxation studies that allowed us to quantitate glycine interactions, hydration and aggregation properties as a function of both pH and amino acid concentration. In the presence of added LiCl to glycine solutions in ethanol-water mixtures (for 60 %

Molecular Modeling - American Chemical Society

Recommend Documents